Robert Flavell: So it's my pleasure today to welcome you all to join us today for this Prostate Cancer Outreach Webinar. This webinar has been organized in collaboration with the Society of Nuclear Medicine and Molecular Imaging Prostate Cancer Outreach Working Group, together with faculty from the University of California in San Francisco. And the topic of the webinar today is on changing the standard of care in prostate cancer with advances in nuclear medicine, PET, and theranostics. So I'll briefly introduce myself to start. So I'm Rob Flavell, I'm an Assistant Professor in the Department of Radiology and Biomedical Imaging at UCSF. Also the Section Chief for Molecular Imaging and Therapeutics. Today we have a great selection of speakers, including both nuclear medicine physicians, as well as clinicians of various different types involved in prostate cancer care. So really clinicians and imagers spanning the spectrum of clinical care in prostate cancer.

Our final speaker today is Dr. Rahul Aggarwal. Dr. Aggarwal is a medical oncologist in the Division of Hematology-Oncology at the University of California in San Francisco. His clinical practice focuses on patients with advanced solid malignancies with a particular emphasis on genitourinary malignancies, including prostate cancer. He's also the leader of the GU medical oncology program at UCSF, and the Associate Director of Clinical Research at UCSF Helen Diller Family Comprehensive Cancer Center. He also has an extensive research program on developing novel therapies and imaging modalities for patients with advanced solid tumor malignancies, the focus on advanced prostate cancer. So thank you very much, Rahul, for agreeing to speak to us today. Thank you very much, Rahul.

Rahul Aggarwal: Great, thanks Rob for the introduction, and thanks everyone for joining. Good afternoon to everyone. Thanks to the organizers for inviting me to speak today. I'm going to give you a little bit of a perspective from a medical oncologist, really focused more in the advanced metastatic prostate cancer setting, sort of moving from oligometastatic, which Felix has really nicely summarized, to perhaps patients with more widely metastatic disease. And now how do we think about both diagnostic and theranostics strategies targeting PSMA? So what does the current evidence show, where the direction's going in terms of where we'd like to see the modality move in the advanced setting?

So there's really three topics I want to cover in a fairly concise fashion. One is what's our current understanding of patterns of PSMA PET expression in castration-resistant disease? We know it's highly sensitive and specific for the detection of disease in the castrate-sensitive setting, including in biochemical recurrence patients with a very low PSA. How does it perform in the disease setting and castration resistance, which has a lot more biologic heterogeneity and loss of differentiation for a certain subset of patients? How do we think about using PSMA PET as a treatment response monitoring tool, getting away from just simple detection, but rather monitoring metabolic response? Is this clinically meaningful? Should we be getting serial PSMA PET scans, if and when this modality becomes FDA approved, should we be using this routinely in our practice as we treat metastatic patients? And then covering what's the lay of the land with theranostics? And certainly, there's been some exciting data presented, including prospective randomized data now that we have from ASCO this year. And so presenting what we currently know, what are the trials out there? And then what are the unanswered questions, things we'd like to see?

So the first question is PSMA expression in mCRPC. We generally regard this as less uniform than in the castration sensitive setting, and probably the best data for this comes from various centers including some of the prospective lutetium PSMA trials where PSMA PET combined with FDG PET was used as a screening tool to screen patients who had to have PSMA positive, FDG negative disease in order to qualify. And clearly, there's a lot of inter and intrapatient heterogeneity with respect to one metastatic lesion versus another. And you can see this as just an example of a patient from this paper that was published back in 2019 showing that yes, clearly there are a lot of PSMA-avid metastatic sites in this patient. However, you also see on FDG PET that there are sites within the liver that show up as metabolically active that do not show up on PSMA. And so the concern that this degree of heterogeneity is that, both from a response monitoring perspective as well as a theranostics perspective, there certainly are lesions, which wouldn't be predicted to respond as well to this type of therapy. So you see this type of heterogeneous picture, not uncommonly. We think of PSMA low or heterogeneous uptake in probably 20, 25% of all mCRPC patients.

Some of which I think this is really due to treatment-emergent neuroendocrine prostate cancer, which we see in a subset of patients. But there's probably a larger degree of mCRPC patients that just simply have heterogeneous disease and lower loss of PSMA expression as the disease advances. These patients that have this kind of discordant uptake, we are starting to believe do have very poor survival outcomes. This is a retrospective study, so certainly lots of caveats, but you can see that the median survival for patients that have this type of FDG high PSMA low subtype or phenotype, the median survival here was less than three months. And so clearly this is a group of patients that we need to pay particular attention to, in a high unmet need to develop new therapies in this setting.

As I alluded to, I think there is a correlation between a loss of PSMA expression and neuroendocrine transformation. I don't think it's the full story, I think there are plenty of tumors that are not histologically defined as neuroendocrine yet still have low PSMA. But I think there is a trend and this is just simple gene expression correlation between FOLH1, which is the gene that produces the PSMA protein, and several neuroendocrine markers, including enolase and NCAM1. There's a paper that was published in 2019 and you see a pretty strong correlation. Clearly, there's a lot of scatter, so this is not a very, very strong correlation, but you do see a trend there.

And then this is just an example of mine, this is a patient that was actually screening for one of our lutetium PSMA trials we have at UCSF. This particular patient, this mCRPC patient, who had been on pretty standard therapies leading up to the screening for the trial, I believe abiraterone and enzalutamide, then had had one or two doses of docetaxel before deciding to stop chemotherapy and screen for the trial. No clinical features necessarily to predict for someone who is going to be low PSMA, but when we did his screening PSMA PET scan, you can see there are clearly hypoattenuating lesions within the liver that looked like metastases, yet have really zero PSMA uptake on the accompanying PET scan. So this was someone who unfortunately screen failed for the trial, was not eligible for lutetium, as we wouldn't think this modality of therapy would benefit him.

So I think in summary with that, we know there's a lot more heterogeneity. I think we're still in the early stages of trying to characterize the patterns of expression of PSMA PET in the metastatic castration-resistant setting. What do the genomics and histology look like in these PSMA low or negative lesions? And can we identify targets for this group of patients?

How about using PSMA PET to monitor treatment response in the metastatic setting as we apply, not just lutetium PSMA, but just standard therapies, including therapies targeting the antigen receptor? Can we use PSMA PET to monitor metabolic response? Early in the treatment phenomenon, we know that there's an inverse relationship between the androgen receptor and PSMA expression. So one might predict that as you start a patient on AR targeted therapy, whether it be primary androgen deprivation in the hormone-sensitive setting or other AR directed therapies in the CRPC setting, would you expect potentially to see actually a flare or early increase in PSMA uptake? And potentially that might be leveraged to enhance expression for subsequent theranostics strategies. I think in reality, although the studies here, these patient numbers are small, we and others have looked at the serial PSMA PET imaging as we start these patients on therapy.

The reality is a lot messier than I think that theory. I think in theory you'd like to see a nice flare, and increased uptake early on. In reality, you see a lot of heterogeneity where some lesions, the lesions show a higher avidity a few weeks after starting, other lesions respond and have decreased uptake, and so you ended up getting a pretty heterogeneous picture. And the exact time point in which this occurs, there's probably temporal heterogeneity as well.

This was just a series of four patients that Tom Hope led this study here, which we all collaborated on. And it's a nice study in that we got serial PSMA PET imaging on a handful of patients as they started on ADT. We also did the same in the CRPC setting with enzalutamide. You can see that some lesions actually do see a flare or increase, yet other lesions show a decrease from baseline. And this is patients, all of which clinically responded to ADT. This is a timeframe of a few weeks after starting therapy.

And then there's been a number of retrospective studies looking at PSMA PET as a response monitoring tool. And these are highlighted in the pink and red shades there in the bottom four rows. Here I think the key difference is that the follow-up PSMA PET scan is further in time. So if you look at the time points here, these are patients who were imaged more like six weeks, or even longer after they started the ADT or other therapies in the CRPC setting. And with a caveat that these are all retrospective, you do start to see a pattern emerge where perhaps the correlation with clinical response as measured by PSA and survival outcomes does seem to correlate fairly well with metabolic response on PSMA PET.

I think there's still a lot of questions to work out in terms of what's the optimal time point to get the follow-up imaging. Does it improve upon the surrogacy that PSA response or PFS might provide, as it relates to OS? As we think about clinical trial design, how do we grade metabolic response? There are certainly some response criteria that have been proposed, those need to be validated. And so luckily PSMA PET is being integrated prospectively into a number of different larger randomized Phase II, Phase III trials. And I think it would be really important to understand the results of those studies before we can just routinely adopt PSMA PET monitoring as a standard response imaging tool. Until we really understand what the data show we shouldn't really be making treatment decisions in the mCRPC setting based on changes on PSMA PET from baseline.

To kind of help answer this question of needing more prospective data, as well as to understand what's the biology of the PSMA PET low or negative lesions, a bunch of us on this call and others, some of our trainees at UCSF have designed a nice prospective study where we're going to be having a companion imaging and biopsy protocol, basically for a lot of our patients that either get started on conventional therapy or clinical trials. Again, not restricted to just lutetium PSMA, but a lot of conventional therapies as well. And we're going to be doing serial PSMA PET imaging, and then trying to enrich our cohort for those patients who have a PSMA low or negative lesion, that's biopsy accessible. Really trying to understand the biology of biopsying these lesions, and seeing what the genomics transcriptional features and histology show.

And then of course looking at correlation with outcomes. How well does PSMA PET perform as a response monitoring tool, and does it correlate with response and survival outcomes? So I think this study type of design is also being employed at other investigational sites as well. And so I think over the next year, 18 months, we hope to have more prospective data to really inform how to use PSMA PET in this setting.

So the third category I'll touch upon today is the theranostics. It's obviously a big category, it's hard to cover in a short talk. There are a number of different PSMA-based theranostics agents with varying forms of radiation, beta particle radiation with lutetium PSMA-617, has probably been the most widely studied because a lot of the retrospective data that came out of Europe and Australia. Peter Mac and Michael Hofman have certainly led a lot of these efforts, and I'll highlight a couple of those studies that they've done. This was the first real prospective high-quality data that we had the lutetium PSMA-617. And of course, with these theranostics agents, the agent binds to PSMA, gets internalized, and then the lutetium particle emits beta particle radiation, which treats the cancer cell.

So this was a Phase II study that actually was pretty small, 30 patients, but was prospective. And that was the key thing that distinguished it from a lot of the data that had come up prior to it. And so they actually did screen patients for eligibility based on that coupling of FDG and PSMA PET, as I alluded to. And you can see the study schema here, patients were treated for up to four cycles, spaced every six weeks apart. You can see that there were two categories of screen failures, those that were just PSMA negative, that's highlighted in the left and number one. And then patient number two is a patient that had some PSMA positive lesions, but there was discordancy and there were clearly lesions that were the higher degree of avidity on FDG PET. Both types of patients were excluded. So when you actually employ that type of stringent criteria, which one would argue is really the proper way to screen patients for lutetium, you end up screen failing 13 out of 43 patients. So it ends up being a non-trivial number that are not eligible.

In terms of the toxicity that we saw that this was again, prospectively followed patients, actually quite manageable, very few patients discontinued for side effects, most stopped for progression. You can see that really, I think just one patient stopped for cytopenias here. You do see some lymphocytopenia and thrombocytopenia, generally recoverable. We haven't seen persistent marrow failure as a significant safety issue yet, although I think we do need to follow patients longer term. Especially as we think about moving these agents into earlier disease setting, where patients may live longer and therefore would be able to follow for a longer period of time to assess the impact on the bone marrow.

Some of the efficacy outcomes from this study, you can see that the response rate was pretty robust. Clearly it's an active agent. You see a 50% or higher decline in 57% of patients. This is a PSA waterfall plot and actually a greater than 80% decline from baseline in 43% of the patients. And so clearly an active drug. I think what's been a little bit underwhelming is the duration of response and progression-free survival, that this was a single-arm study. So of course there was no competitor, but you can see the median PSA progression-free survival was only 6.3 months. Admittedly, this was a pretty heavily pretreated group. The majority had had prior docetaxel and about half had had prior to cabazitaxel chemotherapy. But nevertheless, you're not really seeing a tail on the curve and most patients do ultimately progress. Likewise, for overall median, overall survival was just a shade over 12 months. I think this is important data because it shows that there is definitely room to improve upon lutetium monotherapy. And I think a lot of the efforts are either enhancing PSMA expression or different combination strategies, which I'll touch upon at the end.

This is the Phase III VISION trial. So based on these prospective data, this was the lead registrational trial that has fully accrued, but we don't yet have the results per nct.gov. We do have the primary completion date is currently listed as April of 2021, so we'll see if we get the data by then. But it was a pretty straightforward randomized two to one study of lutetium for up to six cycles, versus best supportive care. Importantly, patients had to have had prior taxane chemotherapy, the supportive care cannot include cabazitaxel. It could include other therapies like AR targeted therapies, but I think that is one of the limitations of this trial, is the comparator arm and whether there's really equipoise in this study. But nevertheless, it did fully enroll and we're awaiting the results of this study.

I think as a follow-up to the Phase III VISION trial, Peter Mac designed and actually reported at ASCO the initial results. We don't have the full results of this randomized Phase II trial. Unlike the VISION trial, that's actually directly compared lutetium to cabazitaxel chemotherapy, these were patients who were post-docetaxel. So I think a much more robust comparator arm for this study. They again used PSMA and FDG PET as a paired screening tool for this study. And again, you had to be PSMA positive without FDG positive, PSMA negative sites on your scan. And so with this type of approach, the screen failure rate was still in that 20 to 30% category. It's a randomized one-to-one study, they looked at PSA response as the primary endpoint. So this was more of a Phase II type endpoint, but I think an important proof of concept study.

You can see here, the screen failure rate as I said was 28%. Again, those two categories of patients, one that are PSMA negative, the other that have discordant FDG positive, PSMA negative lesions. And so that together comprised 28% of all the patients screened for the trial. So I think with this stringent screening, those patients then were subsequently randomized, was 200, and almost equal allocation. There was a pretty high dropout rate in the cabazitaxel arm as an open-label, a fair number of patients who did not want to get the cabazitaxel chemotherapy. So 16 were not treated, of which 15 actually dropped out from this study.

Nevertheless, with the intent to treat analysis, including those patients, you still see an improvement in PSA 50 response. So the 66% with lutetium, very consistent with the prior phase two data, versus 37% with cabazitaxel. Which I think is a reasonable estimate, maybe a little bit of an under-performance of cabazitaxel, but reasonable compared to some of the other data with capacity. So I think it shows clearly there's a potential for added benefit, at least with respect to PSA. We don't yet know about PFS outcomes, a number of events here on the Kaplan-Meier on the right is still smaller than the number required for statistical power. Nevertheless, there was a trend towards the benefit of lutetium. You can see that near the median, the curves kind of come together, but the overall hazard ratio is 0.69. So we'll see how this data plays out with longer follow-up. And of course, looking at longer-term radiographic progression-free survival and overall survival will be important as well.

Clearly, there's a need to understand that despite this enrichment for patients and screening with PSMA and FDG PET, you still are seeing that clearly not every patient's responding, and every cancer's not responding. So what does that due to? Is that the dose of radiation? Is it just the inherent sensitivity of the cancer to radiation? And of course the level of PSMA expression. Not all PSMA positive patients are created equal, there are clearly some lesions that are more avid than others. I think clearly there's a need to really characterize patients, understand the biology, capture biopsies and ctDNA and other potential biomarkers pre and post lutetium, to really understand response.

And this is some of the scoring tools that are really being used to kind of grade the level of PSMA expression. So I think looking at these things, looking at dosimetry and how much lutetium PSMA actually penetrates into the tumor, I think these are going to be important biomarkers as we start to understand who will respond, who doesn't, and who may need other therapies or combination therapies.

So my conclusions, I have two more slides, my conclusions and PSMA uptake in mCRPC, clearly more heterogeneous than earlier stages of the disease. In part maybe due to transformation to neuroendocrine, but I think it's not the whole story. There's probably a broader subset of PSMA low or negative tumors. We do need the prospective imaging data before we can just routinely use PSMA PET to monitor response in the mCRPC setting. So I'm hoping that we get robust data to really inform management decisions and that we're not using PET imaging prematurely and discontinuing patients from therapy, I think is the worry.

Clearly, the theranostics are really promising clear activity, starting to see a signal that it may be superior to second-line chemotherapy, but there are important caveats. We know that number of patients are not eligible, the response rate even in those eligible is not maybe as high, and the response duration maybe not as long as we would hope for. And occasionally toxicity can be an issue, I think we need longer-term follow-up to really understand that.

So where are we going? I think clearly there's interest and already studies that are in planning and being developed to understand or evaluate lutetium PSMA in earlier stages of disease. So chemotherapy-naïve mCRPC, as well as in the castration sensitive setting, and question mark as to whether even earlier. One wonders if at some point the tumor burden is low enough where you wouldn't expect to see necessarily the same magnitude of benefit. But I think it's an open question as to how early this therapy may go, and the data will have to drive us in terms of what we see with the trials.

Understanding the mechanisms of resistance. So those patients that respond, clearly the duration of response as cancers eventually become resistant, and what are the mechanisms that underlie that? How can we enhance PSMA expression? So there's a lot of interesting directions this may go, but hopefully, there are targeted therapies and ways or strategies to increase PSMA expression, to get a low expressing tumor converted, and one that's higher, just straightforward AR targeting therapy, probably not the full answer given the heterogeneity that we see with some of the early PSMA PET data, but hopefully there'll be other strategies that emerge.

A lot of work undergoing with combination strategies, two obvious ones with immunotherapy and PARP inhibitors. Those studies are underway, we're awaiting some of the results of these trials. I think that hopefully there's a path forward for some of these combinations to enhance both the response rate and response duration. Alternative forms of radiation, of course actinium-225 alpha particle admitting compound, PSMA targeting, as well as other surface targets, are in development, potentially a different toxicity profile and shorter particle path. And so it would be interesting to see where that field goes. And then clearly in the PSMA low subsets, are there other targets that we can really develop there for those patients who screen fail for lutetium PSMA? Can we catch them with another type of therapy? Or at the time of resistance if patients have lower PSMA expression as a follow on type of treatment strategy, I think will be important.